AbstractWhen one surveys the work that has been done since the idea of the phoneme first began to take shape — that is to say during a period of more than 70 years — a striking fact emerges, namely that we find no commonly accepted definition of what a phoneme is. Possibly it is indefinable like the fundamental concepts of other sciences (e. g. numbers in mathematics, consciousness in psychology, matter in chemistry). But whether it is definable or not, it is worth noting that three quite different methods of definition have been attempted. Some (including Baudouin de Courtenay, the originator of the term) have regarded it as a mentalistic conception: they suggest that phonemes are abstract sounds which one aims at producing, but which emerge in actual speech as a number of differing concrete sounds depending upon the phonetic environment. Others have regarded it as a physical conception: they express the view that a phoneme is a family of sounds, each of which is appropriate to one or more phonetic contexts, and is unpronounceable in other contexts (in a particular language). Others again maintain that phonemes are phonological conceptions (‘phonological’ in the sense attributed to the term by the Prague School): those features of speech which serve to distinguish one word from another, or ‘utilizable semantic counters’ as some have called them.